Prepared validation for SparseVector indexes - resolves #167 [skip ci]

Co-authored-by: AryanRG <aryanrganjoo@gmail.com>

Author: ankane

pgvector/sparsevec.py

@@ -101,6 +101,8 @@ def _from_dict(self, d: dict[int, float], dim: int) -> None:
         self._dim = int(dim)
         self._indices = [int(v[0]) for v in elements]
         self._values = [float(v[1]) for v in elements]
+        # TODO uncomment in 0.6.0
+        # self._validate()
 
     def _from_sparse(self, arr: sparray | spmatrix, /) -> None:
         value: coo_array | coo_matrix = arr.tocoo(copy=False)  # type: ignore
@@ -138,6 +140,10 @@ def _from_dense(self, value: list[float] | ndarray, /) -> None:
         self._indices = [i for i, v in enumerate(value) if v != 0]
         self._values = [float(value[i]) for i in self._indices]
 
+    def _validate(self):
+        if any(i < 0 or i >= self._dim for i in self._indices):
+            raise ValueError('sparsevec index out of bounds')
+
     @classmethod
     def from_text(cls, value: str, /) -> SparseVector:
         elements, dim = value.split('/', 2)
@@ -171,6 +177,8 @@ def _from_parts(cls, dim: int, indices: list[int], values: list[float], /) -> Sp
         vec._dim = dim
         vec._indices = indices
         vec._values = values
+        # TODO uncomment in 0.6.0
+        # vec._validate()
         return vec
 
     @classmethod

tests/test_sparse_vector.py

@@ -45,6 +45,16 @@ def test_dict(self) -> None:
     def test_dict_empty(self) -> None:
         assert SparseVector({}, 0).to_list() == []
 
+    # TODO uncomment in 0.6.0
+    # def test_dict_negative_index(self) -> None:
+    #     with pytest.raises(ValueError, match='sparsevec index out of bounds'):
+    #         SparseVector({-1: 1}, 3)
+
+    # TODO uncomment in 0.6.0
+    # def test_dict_higher_index(self) -> None:
+    #     with pytest.raises(ValueError, match='sparsevec index out of bounds'):
+    #         SparseVector({3: 1}, 3)
+
     def test_dict_no_dimensions(self) -> None:
         with pytest.raises(ValueError) as error:
             SparseVector({0: 1, 2: 2, 4: 3})  # type: ignore